Thread: [mpls-linux-general] ldp_linux commands
Status: Beta
Brought to you by:
jleu
From: narsi p. <na...@ho...> - 2001-03-13 22:06:57
|
hi, Can some body provide me with some examples for a setup(ingress,egress and a core router)using ldp_linux commands. I would also like to know whether ldp_linux commands have to be used in conjunction with mplsadm commands or can be used independently. thanks in advance narsi _________________________________________________________________ Get your FREE download of MSN Explorer at http://explorer.msn.com |
From: Wenbiao Z. <wz...@cs...> - 2001-03-21 10:52:17
|
Hi, everyone: I am a MSc student who is interested in MPLS, and would like to do my project/thesis in this area. I am new to MPLS, and run into problem when I decide the topics. I would like to receive any useful advice from you. Following is Currently, I have 2 choices for my topic with this software 1. implementation of MPLS Install it on Linux and run it, then get the traffic to do analysis. There are following disadvantage for this option: 1.1 limited LSRs Due to limited resource, I could only have 2 Linux hosts at home (connecting to internet through Cable modem) and 1 or 2 Linux PCs at the school. Since only 3 or 4 LSRs, will the traffic be good enough for analysis? 1.2 do not know what to analysis Even the problem 1.1 does not exist, what kind of analysis could I perform based on the traffic 2. Traffic Engineering In a MPLS seminar by Juniper, a guru suggested me to work on traffic Angering. Due to limited resource, I could only do a little in traffic Angering. For example, in the software, RSVP is not implemented in the software. Once it is implemented, how could I test it to find out it works or not? Before I decide to choose one from above 2 options, would any of you please answer the questions raised in each of the options and give me some advice? Any useful feedback will be highly appreciated. William Zhang School of Computing Science Simon Fraser University Burnaby, BC V5A 1S6 |
From: <sk...@en...> - 2001-03-21 13:36:21
|
> I am a MSc student who is interested in MPLS, > and would like to do my project/thesis in this > area. I'm almost in the same case. I have here at school between 6 and 12 linux PC to do some testing (they are all fitted with 2 ethernet NIC and some have an ATM NIC). I am currently writing a test suite to "bench" routers ; it will generate network flows using tun/tap devices. it will allow to simulate thousands, and up to millions, of network flows with different src/dst ip addresses. I plan to further experiment with LDP, to implement piggybacking over BGP (zebra) and eventually look after freebsd implementation. if anybody is already working on one ot these points, and would like either peer review (well, "newbie review" should be more appropriate) or would think it woud be better to turn my efforts in other directions, feel free to contact me. Jerome Petazzoni <jp...@en...> |
From: James R. L. <jl...@mi...> - 2001-03-21 14:25:49
|
Hello all, First of all, it is very cool that you are both looking to do some research with MPLS, go for it! Second, you should both look into using User Mode Linux to increase the number of LER/LSR you have to work with. UML allows you to run the Linux kernel as a set of user land threads. THus you can run muliple copies of it. I have written an ethernet driver for it, which allows you to network them together. Check out: http://user-mode-linux.sourceforge.net/ NOTE: you will want to grab the latest version of the virtual ethernet driver and probably contact me for help :-) If anyone is interested in how to use this, e-mail the list and I will post a LONG how-to for everyone. Third, Yon has been working on integrating LDP into Zebra. I haven't had a chance to test it yet, but it sounds like he has made lot of progress. Forth, I will be posting a "near 1.0" version of mpls-linux soon. This will hopefully flush out any remaining bugs and provide a stable full feature MPLS implementation for people to play with. THen the real fun begins, working toward 2.0 which will scramble things up quite a bit. I'll post another message with what is planned for 2.0. Fifth, Layer 2 Cross Connects (draft-martini-l2circuit-trans-mpls-05.txt and draft-martini-l2circuit-encap-mpls-01.txt) and MPLS/BGP/VPNS (rfc2548bis) are the services ISP are clamoring to offer. Not much has been done on the lines of researching for these (none that I have seen). Sixth ... I could on for hours. If your looking for more info let me know. Jim On Wed, Mar 21, 2001 at 02:36:15PM +0100, sk...@en... wrote: > > > I am a MSc student who is interested in MPLS, > > and would like to do my project/thesis in this > > area. > > I'm almost in the same case. > > I have here at school between 6 and 12 linux PC to > do some testing (they are all fitted with 2 ethernet > NIC and some have an ATM NIC). > > I am currently writing a test suite to "bench" routers ; > it will generate network flows using tun/tap devices. > it will allow to simulate thousands, and up to millions, > of network flows with different src/dst ip addresses. > > I plan to further experiment with LDP, to implement > piggybacking over BGP (zebra) and eventually look after > freebsd implementation. if anybody is already working > on one ot these points, and would like either peer review > (well, "newbie review" should be more appropriate) or > would think it woud be better to turn my efforts in > other directions, feel free to contact me. > > Jerome Petazzoni <jp...@en...> > > > _______________________________________________ > mpls-linux-general mailing list > mpl...@li... > http://lists.sourceforge.net/lists/listinfo/mpls-linux-general -- James R. Leu |
From: <sk...@en...> - 2001-03-21 14:37:34
|
has anybody thought about embedding MPLS labels in ethernet headers, instead of using a shim ? I thought about using last four bytes of destination MAC address to carry MPLS labels. it would require the receiver to work in promiscuous mode, or to set the multi/broad cast bit of the destination MAC address, or to find some hack in the NIC driver code to allow some kind of filtering. would there be some interest in doing that way, or is there some major pitfall I did not think of ? |
From: James R. L. <jl...@mi...> - 2001-03-21 15:16:24
|
Hello, In te early days of MPLS there was a suggestion to extend the ethernet header to do this, but I don't think the idea ever took. On Wed, Mar 21, 2001 at 03:37:24PM +0100, sk...@en... wrote: > > has anybody thought about embedding MPLS labels in ethernet headers, > instead of using a shim ? I thought about using last four bytes of > destination MAC address to carry MPLS labels. it would require > the receiver to work in promiscuous mode, or to set the multi/broad > cast bit of the destination MAC address, or to find some hack in > the NIC driver code to allow some kind of filtering. You can't really just grab the last 4 octects because the resulting MAC (after overwritting wit the label) might be a valid MAC used by a machine on the network. Jim > > would there be some interest in doing that way, or is there > some major pitfall I did not think of ? > > > _______________________________________________ > mpls-linux-general mailing list > mpl...@li... > http://lists.sourceforge.net/lists/listinfo/mpls-linux-general -- James R. Leu |
From: Sanda D. <dr...@ee...> - 2001-03-29 16:46:37
|
Hi! I am subscribed to: mpl...@li... and I got your mail with the solution for testing MPLS. I am also a Ph.D. student and I am studying MPLS and this solution suits me very well. The problem is that I tried to install uml but I got this error: Kernel panic: VFS: Unable to mount root fs on 62:00 I know that other people had the same problem in August last year and I read the archives from uml mailing list but they didn't helped me. Had anyone the same problem and can anyone help me with this problem? Regards, Sanda On Wed, 21 Mar 2001, James R. Leu wrote: > Hello all, > > First of all, it is very cool that you are both looking to do some research > with MPLS, go for it! > > Second, you should both look into using User Mode Linux to increase the > number of LER/LSR you have to work with. UML allows you to run the Linux > kernel as a set of user land threads. THus you can run muliple copies > of it. I have written an ethernet driver for it, which allows you to > network them together. Check out: > > http://user-mode-linux.sourceforge.net/ > > NOTE: you will want to grab the latest version of the virtual ethernet driver > and probably contact me for help :-) If anyone is interested in how to > use this, e-mail the list and I will post a LONG how-to for everyone. > > Third, Yon has been working on integrating LDP into Zebra. I haven't had a > chance to test it yet, but it sounds like he has made lot of progress. > > Forth, I will be posting a "near 1.0" version of mpls-linux soon. This will > hopefully flush out any remaining bugs and provide a stable full feature > MPLS implementation for people to play with. THen the real fun begins, > working toward 2.0 which will scramble things up quite a bit. I'll post > another message with what is planned for 2.0. > > Fifth, Layer 2 Cross Connects (draft-martini-l2circuit-trans-mpls-05.txt > and draft-martini-l2circuit-encap-mpls-01.txt) and MPLS/BGP/VPNS (rfc2548bis) > are the services ISP are clamoring to offer. Not much has been done on the > lines of researching for these (none that I have seen). > > Sixth ... I could on for hours. If your looking for more info > let me know. > > Jim > > On Wed, Mar 21, 2001 at 02:36:15PM +0100, sk...@en... wrote: > > > > > I am a MSc student who is interested in MPLS, > > > and would like to do my project/thesis in this > > > area. > > > > I'm almost in the same case. > > > > I have here at school between 6 and 12 linux PC to > > do some testing (they are all fitted with 2 ethernet > > NIC and some have an ATM NIC). > > > > I am currently writing a test suite to "bench" routers ; > > it will generate network flows using tun/tap devices. > > it will allow to simulate thousands, and up to millions, > > of network flows with different src/dst ip addresses. > > > > I plan to further experiment with LDP, to implement > > piggybacking over BGP (zebra) and eventually look after > > freebsd implementation. if anybody is already working > > on one ot these points, and would like either peer review > > (well, "newbie review" should be more appropriate) or > > would think it woud be better to turn my efforts in > > other directions, feel free to contact me. > > > > Jerome Petazzoni <jp...@en...> > > > > > > _______________________________________________ > > mpls-linux-general mailing list > > mpl...@li... > > http://lists.sourceforge.net/lists/listinfo/mpls-linux-general > > -- > James R. Leu > > _______________________________________________ > mpls-linux-general mailing list > mpl...@li... > http://lists.sourceforge.net/lists/listinfo/mpls-linux-general > |
From: Yon U. <uk...@rz...> - 2001-03-29 17:04:28
|
On Thu, 29 Mar 2001, Sanda Dragos wrote: [snip] > The problem is that I tried to install uml but I got this error: > > Kernel panic: VFS: Unable to mount root fs on 62:00 make sure your root fs file is named as described in the docs and in the same directory as you are starting uml, a soft link to the real fs-file is enough. HTH, HAND yon |
From: Sanda D. <dr...@ee...> - 2001-03-30 17:40:49
|
Thank's very much for your help. It worked but now it stops right after: ------------------------------------------------------------ ... ... Started device management daemon for /dev Welcome to Red Hat Linux Press 'I' to enter interactive startup Mounting proc filesystem: ------------------------------------------------------------ Can you help me with this one too? Regards, Sanda |
From: Yon U. <uk...@rz...> - 2001-03-31 08:15:24
|
On Fri, 30 Mar 2001, Sanda Dragos wrote: > Thank's very much for your help. It worked but now it stops right > after: > ------------------------------------------------------------ > ... > ... > Started device management daemon for /dev > Welcome to Red Hat Linux > Press 'I' to enter interactive startup > Mounting proc filesystem: > ------------------------------------------------------------ > > Can you help me with this one too? Hmmm,I'm clueless but I can speculate. I've never used RH, only mandrake and then I just as a quick install on an alien box, where I didn't start learning the distro, I just wanted it to work, fast. Which version of RH are you using? I see it is a more or less new one, with devfs support, it might be, that it (devfs support) is not wholly integrated into the system. Try mounting the rootfs on your host system (losetup ...;mount ...) and check out the /dev directory, it should have the usual devices (say ~100 files). Edit the /etc/fstab from the uml rootfs and change the line "/dev devfs ..." to "/devfs devfs ..." and create the /devfs dir in the rootfs-file. Umount, boot uml. If that works, I shall be playing in the lottery, the moon is in my quadrant it seems. Otherwise I'm clueless. greetings, yon |